try to support RECAP as a fragmentation scheme

[UnixJunkie]

[UnixJunkie]

greglandrum left a comment [(greglandrum/rdkit-blog#35)](https://eu-west-1.protection.sophos.com/?d=github.com&u=aHR0cHM6Ly9naXRodWIuY29tL2dyZWdsYW5kcnVtL3Jka2l0LWJsb2cvaXNzdWVzLzM1I2lzc3VlY29tbWVudC0yODgyMTgyMTc2&i=NjZmZjRkMTVjODI1Y2I2YzVhNzBhZjAw&t=aTNuUkZPWHRlQm5xazdOa2FFMlpVQml6dUI0YUhUNEg4ZkhRdldONktpTT0=&h=bc73d92eb0ae4d419a215e5e2599afbd&s=AVNPUEhUT0NFTkNSWVBUSVZKIUddPitBfrZ061Yfq5X8W2ruq3mlFbkg110Jt7aBHQ) 

The RDKit documentation isn't always great, but this is a case where I think it's pretty helpful:

In [11]: help(Recap)
Help on module rdkit.Chem.Recap in rdkit.Chem:

NAME
    rdkit.Chem.Recap - Implementation of the RECAP algorithm from Lewell et al. JCICS *38* 511-522 (1998)

DESCRIPTION
    The published algorithm is implemented more or less without
    modification. The results are returned as a hierarchy of nodes instead
    of just as a set of fragments. The hope is that this will allow a bit
    more flexibility in working with the results.
    
    For example:
    >>> from rdkit import Chem
    >>> from rdkit.Chem import Recap
    >>> m = Chem.MolFromSmiles('C1CC1Oc1ccccc1-c1ncc(OC)cc1')
    >>> res = Recap.RecapDecompose(m)
    >>> res
    <...Chem.Recap.RecapHierarchyNode object at ...>
    >>> sorted(res.children.keys())
    ['*C1CC1', '*c1ccc(OC)cn1', '*c1ccccc1-c1ccc(OC)cn1', '*c1ccccc1OC1CC1']
    >>> sorted(res.GetAllChildren().keys())
    ['*C1CC1', '*c1ccc(OC)cn1', '*c1ccccc1*', '*c1ccccc1-c1ccc(OC)cn1', '*c1ccccc1OC1CC1']
    
    To get the standard set of RECAP results, use GetLeaves():
    >>> leaves=res.GetLeaves()
    >>> sorted(leaves.keys())
    ['*C1CC1', '*c1ccc(OC)cn1', '*c1ccccc1*']
    >>> leaf = leaves['*C1CC1']
    >>> leaf.mol
    <...Chem.rdchem.Mol object at ...>

There's also an example in the "Getting Started" book: [https://www.rdkit.org/docs/GettingStartedInPython.html#recap-implementation](https://eu-west-1.protection.sophos.com/?d=rdkit.org&u=aHR0cHM6Ly93d3cucmRraXQub3JnL2RvY3MvR2V0dGluZ1N0YXJ0ZWRJblB5dGhvbi5odG1sI3JlY2FwLWltcGxlbWVudGF0aW9u&i=NjZmZjRkMTVjODI1Y2I2YzVhNzBhZjAw&t=R2xzQVg4eWxuUWVCTlZxdFU3MUt0ZVdOUUptMGtRMVFKczIxZWxka3VZYz0=&h=bc73d92eb0ae4d419a215e5e2599afbd&s=AVNPUEhUT0NFTkNSWVBUSVZKIUddPitBfrZ061Yfq5X8W2ruq3mlFbkg110Jt7aBHQ)

So you can get the "words" for a molecule like this:

In [3]: m = Chem.MolFromSmiles('CC[NH+](CC)CCOc1ccc(cc1)Cc2ccccc2')
In [7]: h = Recap.RecapDecompose(m)
In [12]: words = list(h.GetLeaves())
In [13]: words
Out[13]: ['*CC[NH+](CC)CC', '*c1ccc(Cc2ccccc2)cc1']